Property price report methodology
Methodology of the dimdom.pl asking-price statistics: the median as the central measure, the 1.5×IQR rule, minimum sample sizes and the data refresh cadence.
This document sets out the methodology behind the asking-price statistics published on dimdom.pl. The empirical basis is the set of active sale and rental listings on the platform; the sections below describe the statistical measures applied, the sample-selection criteria, the outlier-elimination procedure and the known limitations of the dataset.
Asking prices versus transaction prices
The published statistics describe asking prices only, i.e. the amounts declared by sellers at the time a property is exposed to the market. Asking prices differ systematically from transaction prices, which are agreed in the course of negotiations — real-estate market research consistently finds transaction prices below asking prices. The values published in the report should therefore be interpreted as a characterisation of the supply side of the listings market, not as a measure of prices actually realised in transactions.
The median and the outlier-trimmed mean (1.5×IQR)
The primary measure of central tendency adopted in the report is the median price per m² — a positional statistic robust to extreme observations. A complementary arithmetic mean is published after outliers are eliminated according to Tukey's rule: for every observation group (a combination of territorial scope, property type and offer type) the quartiles Q1 and Q3 of the price-per-m² distribution are computed, and observations outside the interval [Q1 − 1.5×(Q3−Q1); Q3 + 1.5×(Q3−Q1)] are excluded from the mean, minimum and maximum. The median is always computed on the full set of observations; the number of excluded observations is reported in the published data files.
Minimum sample size for publication (n ≥ 30)
Numerical statistics are published only for groups comprising at least 30 active listings. Below this threshold, estimators of central tendency are too unstable to support reliable inference, so an insufficient-sample annotation is displayed in place of numerical values.
Minimum data-presence threshold (n ≥ 3)
Independently of the publication threshold, a given combination of location and property type is recorded in the output dataset only when at least 3 active listings exist. This limits the risk of indirectly identifying individual listings in locations with very few offers.
Refresh cadence
Data aggregation runs automatically once per day, at approximately 03:20 Central European Time (Europe/Warsaw), on the set of listings active on the platform at computation time.
Construction of the monthly series
The points of the time series shown on the charts correspond to the median of the daily readings within a calendar month, rather than to any single reading. This construction limits the influence of short-term fluctuations in the number and structure of listings on the series.